fntnul •/ Ptn—ittty mi Ucial PtyckoUa 1974. Vol. M. No.

J, 364-3 7S Basking in Reflected Glory: Three

(Football) Field Studies Robert B. Cialdini Arizona State University Avril Thorne Arizona State University
Richard J. Borden Purdue University Marcus Randall Walker and Stephen Freeman Ohio State University
Lloyd Reynolds Sloan University of Notre Dame

The tendency to "bask in reflected glory" (BIRG) by publicly announcing one's associations with successful
others was investigated in three field experiments. All three studies showed this effect to occur even
though the person striving to bask in the glory of a successful source was not involved in the cause of the
source's success. Experiment 1 demonstrated the BIRG phenomenon by showing a greater tendency for
university students to wear schoolidentifying apparel after their school's football team had been
victorious than nonvictorious. Experiments 2 and 3 replicated this effect by showing that students used
the pronoun tee more when describing a victory than a nonvktory of their school's football team. A model
was developed asserting that the BIRG response represents an attempt to enhance one's public image.
Experiments 2 and 3 indicated, in support of this assertion, that the tendency to proclaim a connection
with a positive source was strongest when one's publk image was threatened. It is a common and
understandable tendency for people who have been successful in some positive way to make others
aware of their connection with that accomplishment. However, there also appears to be a seemingly less
rational but perhaps more interesting tendency for people to publicize a connection with another person
who has been successful. This latter inclination might be called the tendency to bask in reflected glory
(BIRG). That is, people appear to feel that they can share in the glory of a successful other with whom they
are in some way associated; one manifestation of this feeling is the public trumpeting of the association.
Such a phenomenon is not hard to understand when the one wishing to share in another's success has
been instrumental to that success. However, the more intriguing form of the phenomenon occurs when
the one who basks in the glory of another has done nothing to Requests for reprints should be sent to
Robert B. Ctaidini, Department of Psychology, Arizona State University, Tempe, Arizona 85281. bring
about the other's success. Here, i simple case of affiliation or membership U sufficient to stimulate a public
announcement of the critical connection. There does seem to be abundant anecdotal evidence that
people try to make us cognizant of their connections with highly positive or successful others. The forms
of these connections are varied. For example, they may imply similarity of residence, past or present
States and cities like to list the names of famous entertainers, statesmen, beauty contest winners, etc.,
who live or were born within their boundaries; the state of Indiani has even gone so far as to brag that
mor* vice-presidents of the United States have come from Indiana than any other state. Other such
connections involve ethnic or religious affiliation: Italians speak proudly of the ethnic background of
Marconi, and Jew; refer to Einstein's heritage. Still other connections reflect physical similarities:
"Napoleon was short, too." Sexual identity may also give rise to the BIRG phenomenon: At .t women's
movement forum attended by on: of the authors, there was a round of feminin; 366 BASKING IN
REFLECTED GLORY 367 applause when it was announced that Madame Curie was a woman and Lee Harvey
Oswald was not. Finally, connections suitable for BIRGing may be as tenuous as an incidental contact: We
all know people who delight in recounting the time they were in the same theater, airplane, or restroom
with a famous movie star. While there appears to be rich informal support of the sort described above for
the existence of a BIRG phenomenon, there seem to be no experimental investigations of the effect. Thus,
it was the purpose of this series of studies to examine this tendency to bask in the reflected glory of
another or group of others. In so doing, it was hoped to (a) reliably demonstrate the existence of the
phenomenon, (b) establish its generality over experimental contexts and measures, (c) determine a
mediating process for its occurrence, and (d) discover some of its limiting conditions and thereby gain
further information as to its nature. One of the most obvious arenas for the working of BIRG effects in our
society is the athletic arena. Fans of championship teams gloat over their team's accomplishments and
proclaim their affiliation with buttons on their clothes, bumper stickers on their cars, and banners on their
public buildings. Despite the fact that they have never caught a ball or thrown a block in support of their
team's success, the tendency of such fans is to claim for themselves part of the team's glory; it is perhaps
informative that the chant is always "We're number one," never "They're number one." It was our view
that a sports context would be ideal for a test of some of our notions concerning BIRG effects. Our
expectation was that an individual would attempt to bask in the glory of an associated, successful source
by publicly announcing his or her affiliation with the source and that this effect would obtain even when
the affiliation was clearly irrelevant (i.e., noninstrumental) to the success of the source. In order to gather
data relevant to the above hypothesis, an experiment was simultaneously conducted at seven universities
with powerful intercollegiate football teams during part of the 1973 football season. It was predicted that
students at these schools would be more likely to announce publicly their connection with their
universities after the varsity football teams had been successful than after the teams had not been
successful. We decided to measure students' tendency to announce their university affiliation by means
of an examination of wearing apparel. The frequency with which students wore apparel that clearly
identified the university that they attended was hoped to be a subtle yet sensitive measure of the
willingness to declare publicly a university affiliation. EXPERIMENT 1 Method Procedure. From the third
week of the 1973 collegiate football season through the last week of regular play, the apparel of students
enrolled in sections of introductory psychology courses at seven large universities was covertly monitored.
At each school, three types of data were recorded in the same classes every Monday during the season:
(a) the number of students present in the class, (b) the number of students with apparel identifying the
school of attendance, and (c) the number of students with apparel identifying a school other than the
school of attendance. Data recorders at each place received the following definitions prior to data
collection: Apparel identifying the school of attendance b identified as apparel which unambiguously
identifies your school through names, insignia, or emblems. Examples would be buttons, jackets,
sweatshirts, tee shirts, etc., which display the school name, team nickname or mascot, or university
insignia. Apparel which appears school-related solely through the use of colors would not quality. Also
excluded are utilitarian objects which announce a university affiliation such as briefcase*, notebooks, or
bookcovers. Apparel identifying a school other than the school of attendance are those which meet the
same criteria for inclusion as above but which identify a school other than your own. The data recorders
were not members of the classes they monitored. Results Over all schools and across all weeks, an average
of 176.8 students were present in the monitored classes; an average of 8.4% of these students wore
apparel identifying the university of attendance, while Ifo of &eflt 368 CIALDINI, BORDEN, THORNE,
WALKER, FREEMAN, AND SLOAN wore apparel identifying a school other than the university of
attendance. Because of huge differences among the schools in absolute amounts of these two kinds of
apparel wearing and in order to make comparisons between the universities as well as between the types
of apparel wearing, standardized indexes of relevant apparel wearing were considered necessary. The
standard we decided on was the highest percentage of relevant apparel wearing that occurred on any
Monday during the season; this standard was simply computed as the number of students wearing
relevant apparel that day divided by the number of students in class that day. The percentages of apparel
wearing on all other Mondays of the season were scored as proportions of the highest percentage. So,
the Monday with the largest percentage of relevant apparel wearing was scored as 1.00, and any other
Monday percentage was scored as a fraction (proportion) of that standard. This procedure was performed
on the data from each school for the two relevant categories of apparel wearing: school-of-attendance
apparel wearing and school-of-nonattendance apparel wearing. A mean proportion for each category was
obtained for Mondays following a team's wins and nonwins; these are the mean proportions presented
in Table 1. As can be seen from Table 1, these indexes showed a generally consistent tendency for students
to wear school-of-attendance apparel more after vicTABLE 1 INDEXES or RELEVANT APPAREL WEARING
SEVEN MONITORED UNIVERSITIES School Arizona State Louisiana State Ohio State Notre Dame MlfhigHn
Pittsburgh Sototbern California M School-ofattendance apparel wearing Wins .63(5) .80 (5) .69(4) •67 (7)
.52 (5) .76 (4) .36 (6) 63 Nonwins •61 (1) .33 (3) .30(1) •49 (1) .83(1) •27 (2) 26(1) 44 - AT THE School-
ofnonattemtance apparel wearing Wins .58 (5) .58 (5) .56(4) .62 (7) .20 (5) .31 (4) 17(6) 43 Nonwins •68(1)
.5! (3) •94(1) .52 (1) .00(1) .50(2) .ooo) Nltt, Numbers In parentheses represent the number of games that
/ell iota wins and nonwins categories (or each school. tones than nonvictories; but this was not the case
for school-of-nonattendance apparel. Because of the non-normality of the proportion data, the scores
were converted to ranks, and Wilcoxon matched-pairs signed-ranks tests were performed using school as
the unit of analysis.1 Despite the conservativeness of such an approach (for this mode of analysis, n is only
7), the Wilcoxon T reflected a conventionally significant difference on the schoolof-attendance measure
(T = 2, p < .OS, twotailed). This result indicated, as predicted, that Mondays following football victories
ranked significantly higher in school-of-attendance apparel wearing than Mondays following
nonvictories.2 The mean rank for victories was 3.2, while that for nonvictories was 4.Q. A similar test for
school-of-nonattendance apparel did not show any effect; the victory and nonvictory mean ranks for this
measure were 3.4 and 3.7, respectively. This latter result suggests that the obtained effect on the school-
of-attendance measure is not attributable to a simple tendency to wear clothing of a certain type (e.g.,
athletic team jackets, sweat shirts, tee shirts, etc.) after an athletic team victory. Discussion In all, we
found support for our expectations concerning the BIRG phenomenon. 1 In any conversion of parametric
data to ranks, the possibility exists that the ranked scores will not fully reflect the character of the
parametric data. In order to examine such a possibility with respect to our results, a correlational analysis
was performed on the standardized index scores and their derived rank scores. A highly similar
relationship (r — —.8.1) between the two forma of scores was found; the negativity of this correlation is
simply due to the fact that the better ranks of those of lower numerical value. 2 It may be instructive to
note that the single exception from this pattern in Table 1 occurred at the University of Michigan as a
direct result of a 10-10 tie with Ohio State University in a game for the Big Ten Conference Championship.
Most observers, especially the Michigan supporters, felt that the Michigan team had outplayed Ohio State
that day and that the game demonstrated Michigan's superiority. However, that tie game constituted
Michigan's only entry in our nonwin category, resulting in the only reversal in our data. BASKING IN
REFLECTED GLORY 369 Students chose to display more apparel indiators of their academic affiliation after
their jniversity's varsity football team had re:ently been successful. It appears, then, from these data and
from numerous anecdotal rei>orts that people desire to make others aware if what seem to be their
causally meaning'ess associations with positive sources.8 Why? what do they intend to get from it?
Perhaps, the answer has to do with Heider's balance formulation (19S8). Heider discussed two types of
perceived relations between things: sentiment relations, which imply a feeling state between stimuli, and
unit relations, which merely imply that things are connected in some manner. It is the unit relationship
that seems akin to the noninstrumental connection that people tend to publicize between themselves
and a successful or otherwise positive source. The results of the present experiment could well be seen
as consistent with balance theory. For example, if observers perceive a positive unit relationship (e.g.,
university affiliation) between a student and a successful football team and if observers generally evaluate
successful teams positively, then in order to keep their cognitive systems in balance, the observers would
have to evaluate the student positively as well. Hence, we might expect the student to want to make the
unit connection evident to as many observers as possible, in this case, through the wearing of university-
identifying clothing. The process whereby one publicly seeks to associate himself or herself with a
successful other, then, may be reinforced by the tendency of observers to respond in a similar fashion to
associated stimuli. Indirect evidence that tends to support this hypothesis comes from research
concerning the transmission of positive and negative information. Manis, Cornell, and Moore (1974) have
shown that one who transmits information that the recipient favors is liked more by the recipient than
one who transmits information that the recipient disfavors and that this liking occurred even though it
was understood that the transmitters did not necessarily endorse the communicated information. Like
the royal messengers of old Persia who were feted when they brought news of military victory but killed
when they brought news of defeat, the transmitters in the Manis et al. (1974) study acquired the valence
of the message with which they were simply paired. Moreover, there is evidence that people recognize
this generalization effect and tend to take actions that connect them, in the eyes of observers, with
positive rather than negative news. For example, Rosen and Tesser have repeatedly shown (e.g., Rosen &
Tesser, 1970; Tesser, Rosen, & Batchelor, 1972; Tesser, Rosen, & Tesser, 1971) that people prefer to be
connected with the communication of good news to another than with the communication of bad news.
Investigating the basic effect, Johnson, Conlee, and Tesser (1974) found their subjects reluctant to
communicate negative information not because they felt guilty about transmitting bad news but because
they feared that they would be negatively evaluated by the recipient of such news; again, this was true
even though all concerned knew that the communicators had in no way caused the bad news. Thus, it
appears from these data that: first, individuals who are merely associated with a positive or negative
stimulus (in this case, favorable or unfavorable information) will tend to share, in an observer's eyes, the
affective quality of the stimulus; and second, at some level individuals seem to understand the workings
of this phenomenon and make use of it in the ways they present themselves to others. We wish to
interpret the results of Experiment 1 in terms of this formulation. Students at our seven monitored
universities chose to wear school-of-attendance apparel after football team victories in order to display
their connection with the successful team and thereby to enhance their esteem in the eyes of observers
to the connection. However, another explanation of our findings exists as welL Perhaps the tendency to
wear university8 It might be argued that some subjects felt that their presence in the stands on the day
of a game directly contributed to their team's success. This seems an unlikely explanation for the obtained
results, as an analysis of the data of Experiment 1 showed an equally strong BIRG effect for home • away
games. 370 CIALDBH, BORDEN, THORNE, WALKER, FREEMAN, AND SLOAN related clothing following team
wins had nothing to do with an attempt to proclaim the favorable connection to others but only reflected
an increased positivity toward the university as a consequence of team success. That is, it is possible that
a football victory caused students to like their school more, and this heightened attraction manifested
itself in the tendency to wear school-identified apparel. To test these alternative explanations and to
establish the generality of the tendency to BIRG in a different experimental situation than that of
Experiment 1, a second experiment was conducted. EXPERIMENT 2 The major distinction between the
competing interpretations described above is the contention of the BIRG model that students wore
school-of-attendance clothing after victories in order to publicize their university affiliations and hence
increase their prestige in the view of others. The "heightened attraction" formulation makes no such
claim: (tee is simply seen to like the school more following victories, and this, rather than the possibility
of increased interpersonal prestige, is said to stimulate the wearing of relevant apparel. We decided to
test these explanations by way of an examination of the pronoun usage of university students describing
the outcome of one of their school's football contests. Earlier in this article we alluded to the tendency of
athletic fans to crowd in front of television cameras, wave their index fingers high, and shout, "We're
number one!" The choice of this pronoun seemed to us a very good measure of the tendency to BIRG. By
employing the pronoun we, one is publicly able to associate oneself with another person or group of
persons. Through the use of some other designation, for example, they, one is able to distance oneself
from (i.e., to weaken the perceived association with) another person or persons. It was our feeling that in
order to BIRG a successful football team, students would be more likely to describe the outcome of a team
victory using the pronoun we than they would a team nonvictory. Thus, it was our expectation that this
tendency to connect oneself with a positive source but distance oneself from a negative source would
influence subjects to use the term, "We won," to describe a team win but use the third person (e.g., "They
lost") to describe a team loss.4 Further, in line with our BIRG model, it was expected that this differential
use of language would be most pronounced when the subject's esteem in the eyes of an observer had
been recently lowered. That is, if we are correct in proposing that one proclaims a connection with a
positive source in an attempt to raise one's esteem in the view of others, then one should be most likely
to declare such a connection when that esteem has recently been jeopardized. Thus, if we were to create
experimentally in subjects a need to bolster esteem in the eyes of an observer, subjects should be most
likely to announce publicly (through use of the pronoun we) a connection with a successful team and be
least likely to publicize a connection with an unsuccessful team. On the other hand, subjects who have
less need to elevate an observer's evaluation of themselves should show a lesser effect. The simple
"heightened attraction" model would not make such a prediction, since one's prestige in the eyes of
others is not a critical variable in that formulation. Method Subjects. The subjects were 173
undergraduates at a large state university with a nationally ranked football team. Subjects were randomly
selected from student listings in the university phone directory. The sample included approximately equal
distributions of males and females. Procedure. During a 3-day period midway through the 1974 football
season, subjects were contacted or the phone by one of 16 experimenters (eight male; and eight females)
identified as an employee of a "Regional Survey Center" with headquarters in ar out-of-state city. The
caller explained that he (she) was conducting a survey of college students' knowledge of campus issues
and was in town that da; calling students at the subject's university. Subject' agreeing to participate (93%)
were then asked s series of six factually oriented forced-choice questions about aspects of campus life
(e.g., "What per cent of students at your school are married? Wouk you say it's closer to 20% or 35%?").
Followinf * It should be evident to the reader that th< general statement of the BIRG formulation include'
not only the tendency to bask in reflected glory bu also the tendency to distance unattractive sources.
BASKING IN REFLECTED GLORY 371 the subject's sixth response, the caller administered the first
manipulation. Half of the subjects were told that they had done well on the test, and half were told that
they had done poorly. Specifically, subjects were told: That completes the first part of the questions. The
average student gets three out of six correct. You got [five; one] out of six correct. That means you [did
really well; didn't do so well] compared to the average student. Subjects were then to'.d that there were
a few more questions and that the first concerned students' knowledge of campus athletic events. At this
point the second experimental manipulation occurred. Half of the subjects were asked to describe the
outcome of a specific football game; their school's football team had won this game. The other half were
asked to describe the outcome of a different game; this was a game that their team had lost. The question
was phrased as follows: In the [first; third] game of the season, your school's football team played the
University of [Houston, Missouri]. Can you tell me the outcome of that game? If a subject did not know
the results of the game, a new subject was called. Otherwise the subject's verbatim description of the
game outcome was recorded. At the end of the interview, all subjects were fully debriefed. Independent
variables. Two factors were manipulated; a subject's personal outcome on the survey task (success or
failure) and the affiliated football team's outcome in the game described (win or nonwin). These factors
combined to produce a 2 X 2 factorial design. Dependent variables. Subjects' tendency to use a we or non-
we response in describing a team outcome constituted our dependent measure. Descriptions such as "We
won," "We got beat," etc., were considered we responses. All other descriptions (e.g., "The score was 14-
6, Missouri." "They lost.") were classified as non-we responses. Predictions. Two predictions were made.
First, it was hypothesized that subjects would emit more we responses in describing a team victory than
a team defeat. Second, it was expected that the effect of Hypothesis 1 would be greatest for subjects who
had "failed" the survey test. The latter hypothesis was based on the assumption that subjects in the
personal failure conditions would attempt to associate themselves publicly with a positive event or
'tistance themselves from a negative event through language usage in order to bolster or salvage their
damaged images in the eyes of the caller. Subjects in the personal success conditions were not expected
to show a similar sized effect, as their prestige had already been ensured via their successful task
performance. Evidence that public success and failure on an experimental task leads to differential
tendencies for social approval has been offered by Schneider (1969). He manipulated success and failure
and found failure subjects to present themselves more favorably to an observer who could provide an
evaluation. Thus, if the BIRG phenomenon ii indeed an attempt to gain social approval, we should see our
failure subjects BIRG more than our succesi subjects. Results Of the 173 subjects, the data of 5 were
discarded because they clearly reported the game results incorrectly. For example, the description "We
won" was not counted if io fact the subject's team had lost the game in question. The percentages of we
responses emitted in the four experimental conditions are presented in Table 2. The first prediction, that
we usage would be greater in the descriptions of team victories than team defeats, was tested by
comparing the team win conditions against the team nonwin conditions. A significant effect was obtained,
x iU ) = 4«20 , p < .05, confirming Hypothesis 1. The second prediction, that the tendency for we responses
to attend victory rather than defeat descriptions would be strongest after a personal failure, was tested
as an interaction of the two major independent variables. The resultant statistic, suggested by Langer and
Abelson (1972) for testing interactions within a 2 X 2 contingency table, just missed conventional
significance levels, Z = 1.75, p < .08, two-tailed. Tests of the simple main effects of the interaction strongly
supported Hypothesis 2. The difference in we responding between the team success and team failure cells
of the personal failure condition was highly significant, x 2O) = 6.90, p < .01. The comparable test within
the personal success condition did not approach significance, x*(l) = .07, ns. There were no significant sex
effects in the data. TABLE 2 PEBCENTAGE OF SUBJECTS USING "WJS", EXPERIMENT 2 Team outcome Win
Nonwin % Personal outcome Succtse 24 (11/45) 22 (9/41) Failure 40 (16/40) 14 (6/42) Mean % 32(27/88
» 18 (15/a3) 372 CIALD1NI, BORDEN, THORNE, WALKER, FREEMAN, AND SLOAN DISCUSSION The data of
Experiment 2 seem clearly to support the general BIRG formulation. Subjects used the pronoun we to
associate themselves more with a positive than a negative source, and this effect was most pronounced
when their public prestige was in jeopardy. We interpret these results as evidence for our contention that
people display even the most noninstrumental connections between themselves and the success of others
so as to receive positive evaluations from the observers of those connections. It should be evident that
the observer's tendency to assign positivity to one who is associated with positive things is crucial to our
hypothesizing about the BIRG phenomenon. It follows from our previously stated assumption that if a
person understood that a given observer did not value the success of a specific source, that person would
be less likely to try to BIRG that source to the observer. So, if one of our subjects knew that an observer
abhored successful college athletic programs, we would predict that there would be little likelihood of the
subject attempting to make visible a connection with a winning football team. But this is a fairly obvious
example; few people would predict otherwise. A more subtle and perhaps more informative
demonstration might be obtained through a somewhat different manipulation of the observer's
relationship to the connection. When an observer to a highly positive association can also lay claim to the
association, the prestige of the connection is diffused and, consequently, reduced for anyone attempting
to bask in its glory. It is when one's bond to a positive source is not shared by an audience that its prestige
value is optimal. Thus, when everyone has a similar positive characteristic, there is no special prestige
involved in possessing it. and the likelihood that any one person will boast about that quality should be
reduced. For example, a resident of California is less likely to brag to fellow Californians about the
favorable climate than to geographically distant others, especially those who cannot claim similarly
pleasant weather. It is our hypothesis, then, that the tendency to BIRG a positive source should occur
most often when one's connection with the source is stronger than the observer's." A third study was
conducted to test this contention. EXPERIMENT 3 In Experiment 2, it was shown that a personal failure
experience increased our subjects' tendency to associate themselves with a positive source and decreased
their tendency to associate themselves with a negative source. We have argued that this result occurred
because the failure experience lowered perceived prestige and motivated subjects to try to either bolster
their images in the eyes of others or prevent them from being further degraded. Central to this argument
is the assumption that one's simple, noninstrumental connections are seen to influence observers'
personal evaluations. If so, it should be the case that in addition to their use as dependent measures, such
connections could be used as effective independent variables. That is, it should be possible to influence
subjects' behavior by publicly connecting them with either positive or negative events. In fact, if we are
correct in our assumption, manipulating one's public connections with good or bad things should have
the same effect as manipulating one's personal success or failure experiences. For example, just as
Experiment 2 showed that subjects who failed a task increased the tendency to affiliate themselves with
a winner and decreased the tendency to affiliate themselves with a loser in the eyes of an observer to
their failure, it follows from our formulation that subjects who are merely publicly connected with a
negative event should emit comparable BIRG responses in the presence of an observer to that connection.
Experiment 3 was designed to test this possibility and represented a conceptual replication and extension
of Experiment 2. 6 We do not wish to suggest that the tendency tc BIRG a positive source never takes
place when thi observer's association to a successful other is a strong as one's own but only that the
prestige to lx derived from a unique (vis-a-vis the observer) connection is relatively more desirable.
BASKING IN REFLECTED GLORY 373 $ cthod Subjects. The subjects were 170 undergraduates at a large
state university with a powerful football t .m. The university was not the same as that of r perlment 2;
however, subjects were selected for p rtldpation in a fashion identical to that of I vperiment 2. Procedure.
Following the completion of play for t .• university's football team, subjects were called r the phone by 1
of 18 experimenters (11 males . d 7 females) identified as an employee of either i _• "university's Survey
Center" located on campus i the "Regional Survey Center" located in an outi state city. Subjects were told
that a survey was I mg conducted of "undergraduates' knowledge of i iversity athletic events." Those
agreeing to parI ipate (96%) were asked to describe the outcome o' first one, then another of their football
team's 1., t games of the year. One of the games constituted ai important victory, and the other an
important n nvictory in the team's season. Half of the subjects v re first requested to describe the
nonvictory game ,i 1, having responded, to describe the victory game. 1 e other subjects had the requests
put to them in reverse order. The subjects' verbatim descriptions 0' the game results were recorded.
Independent variables. Two factors were orthogon.ly varied: the strength of the subject's affiliation to the
university team compared with that of the ol' erver (same as observer's or stronger than obsi ver's) and
order of presentation of the games to b> described (victory game description requested fi' t or nonvictory
game description requested first). dependent variables. The dependent measure was tt pattern of toe and
non-voe usage employed by snjects to describe the combination of the victory ai 1 nonvictory games.
Three combinations were possn le. A subject could have used the same we or ni i-v/e term to describe
both the victory and the rn ivictory, could have used vie to describe the nonv 'ory, and non-we to describe
the victory, or fi illy, could have used we for a victory and non-we fi' a nonvictory. 'rediclions. It was
predicted, first, that there v. uld be an overall tendency for subjects to use a in their descriptions of a team
victory and n< t-we in their descriptions of a team nonvictory. S h a finding would replicate the basic BIRG
c! ct obtained in Studies 1 and 2. A second hypothec- was that the tendency to use we for victory and n -
we for nonvictory descriptions would be greater it the nonvictory-description-requested-first conditi is.
Such a result would constitute a conceptual r> lication of the second finding of Experiment 2. 0 the basis
of the BIRG model, we expected that tt effect of publicly describing a negative event w h which one is
connected would be equivalent in n are to publicly failing on a task. Both operations » e thought to reduce
subjects perceptions of their p >tige as seen by an observer and, hence, to inci ise the likelihood of
subjects' attempts to ensure tl positivity of subsequent evaluations. The third prediction was that
Hypothesis 2 would hold most strongly when the observer was identified with an off-campus organization.
This expectation was based on the belief that felt prestige to be gained from one's connections to a source
is greater and, thus more sought after, when one's connections to that source are stronger than an
observer's. Confirmation of this prediction would appear as an interaction of the independent variables
of the study.* Results As expected and consistent with the results of Experiments 1 and 2, the basic BIRG
effect occurred in Experiment 3 to support our first hypothesis. That is, subjects used the term we nearly
twice as often to describe a victory than a nonvictory (26% vs. 13.5%). This effect is further confirmed
when the data are examined in terms of individual subjects' we/non-we usage patterns. The majority of
subjects were constant in their pattern of responding to the two requests for descriptions; they
consistently used either we or non-we to describe both game outcomes. Thus, there was a strong
tendency for our subjects to be consistent in their verbal usage patterns for the two descriptions.
However, in 23 instances subjects provided an inconsistent we/non-we pattern. In 22 of those instances,
the pattern supported the BIRG model; the pronoun we was used for the victory description and a non-
we term was used for the nonvictory description. Using McNemar's test for the significance of changes
(Siegel, 1956, pp. 63-67) 6 The experimenters, undergraduate students in a laboratory social psychology
course, were not fully aware of these predictions. In order to test the influence of conscious or
unconscious experimenter bias on the results of this study, the experimenters were informed of the
nature of Hypothesis 1. However, they were blind to the more subtle Hypotheses 2 and 3. If only
Hypothesis 1 were confirmed, the data would likely have to be interpreted as potentially influenced by
the experimenter bias artifact. The investigation of such a possibility was deemed an important one, since
in the prior experiments, experimenters had knowledge of the experimental hypothesis. In Experiment 1,
some data recorders were unintentionally informed of the major hypothesis, while others were not. An
analysis of the data from these two groups found only a minimal difference in the data patterns, with the
uninformed group's data actually more favorable to prediction than the informed group. However, in
Experiment 2, all experimenters had knowledge of the prediction. 374 CIALDINI, BORDEN, THORNE,
WALKER, FREEMAN, AND SLOAN TABLE 3 PERCENTAGE OF SUBJECTS USINO BOTH "W V FOR VICTORY
DESCRIPTIONS AND "NON-WE" FOR NONVICTORY DESCRIPTIONS, EXPERIMENT 3 Order of request'
Strength of subject's connection to team relative to observer's Victory description requested 6rst
Nonvlctory description requested first Stronger than observer's Same as observer's Mean% 3 (1/39) 21
(10/47) 11 (4/36) 14 ( 7/4S) 7 (5/75) 18 (17/95) * Numbers given are percentages and correcting for
continuity, the data are highly significant, X J (1) = 17.39, p < .001. The tests of Hypothesis 2 and 3 were
conducted by considering the distribution (across the cells of our design) of the instances of we/non-we
usage fitting the pattern predicted by the BIRG model. Table 3 presents these data. Hypothesis 2 stated
that more subjects would use we to describe the victory and non-we to describe the nonvictory when
they were asked to describe the nonvictory first. As expected, subjects were significantly more likely to so
respond in the nonvictory-description-requested-first conditions, x*(l)=4.69, P < .05. Hypothesis 3 stated
that Hypothesis 2 would hold most clearly when the subjects were more strongly connected with the
university team than was the observer. As predicted, Hypothesis 2 was supported to a greater extent
when the observer was affiliated with an off-campus rather than a campus agency. However, this
tendency did not quite reach conventional levels of significance, Z = 1.72, p < .085. As in Experiment 2,
there were no significant effects for sex of subject. GENERAL DISCUSSION Overall, Experiments 1,2, and 3
provided strong support for the BIRG formulation. All three experiments showed a significant tendency
for students to strive to associate themselves publicly with their university's football team more after the
team had been successful. A striking aspect of the phenomenon is that subjects sought to proclaim their
affiliation with a successful source even when they in no way caused the source's success. This component
of the effect suggests a mediator consistent with balance theory. It is our contention that people make
known their noninstrumental connections with positive sources because they understand that observers
to these connections tend to evaluate connected objects similarly. It appears that the tendency to BIRG
is an attempt to secure esteem from those who can perceive the cornection. Studies 2 and 3 provided
support fc r such an interpretation. Both showed that e\- perimental operations designed to threaten z
subject's esteem in the eyes of an observu caused subjects to be more likely to try pul>- licly to associate
themselves with posithe sources. Intriguingly, it was possible to increase the tendency to BIRG in these
experiments either by initially causing the subject to experience personal failure in an observer „ eyes or
by initially causing the subject to be noninstrumentally connected with a negathe event in an observer's
eyes. These manipulations proved functionally equivalent in modifying subject pronoun usage. Thus, in
support of our basic argument, being merely associated with someone else's success and failure had much
the same effect as personal succe^ and failure. Experiment 3 provided evidence in a different way as well
that the desire for prestige is the mediator of the BIRG response. It demonstrated that when subjects'
affiliation with a positive source was stronger than an observer's (and therefore carried a greater amount
of prestige), they were most likely to BIRG that source in the presence of the observer. These studies
suggest a way to understand how the fortunes of affiliated sports teams cm cause lavish displays of civic
gratitude and pride in American cities, or "sports riots" in Europe, or murders in South America of players
and referees whose actions had caused a home-team defeat. Through their simrle connections with sports
teams, the persona! images of fans are at stake when their tear is take the field. The team's victories and
cefeats are reacted to as personal successes a id failures. Throughout this article we have stressed an
interpersonal mediator of the BIRG phenomenon—the perceived esteem of others. V BASKING IN
REFI.ECTED GLORY 375 do not wish, however, to preclude the possibility of the tendency to BIRG privately.
That is, for wholly intrapersonal reasons, people may draw connections between themselves and positive
sources. For example, one may well feel an enhancement of self-esteem that is unrelated to the
assessments of others when one is associated with success or positivity. Such an effect could also be
interpreted in terms of a tendency to respond similarly to associated objects. It might be that the results
of our experiments are, in some degree at least, due to a desire to bolster or maintain one's self-concept.
The tendency to employ appropriate apparel or language in a way that connects oneself to something
good may involve an attempt to remind oneself of such connections and, thereby, positively affect self-
esteem. The fact that in Experiment 2 we were able to influence the BIRG response simply by manipulating
the characteristics of the observer suggests that the BIRG phenomenon is not mediated solely by
intrapersonal phenomena. Nonetheless, it remains iwssible that the tendency to BIRG has its basis in a
desire to affect self-image as well as social image. In fact, since there is evidence that how we regard
ourselves is influenced by how we perceive that others regard us (e.g., Harvey, Kelley, & Shapiro, 1957),
these two mediators are not mutually exclusive. REFERENCES Harvey, 0 . J., Kelley, H. H., & Shapiro, M. M.
Reactions to un(a\orable evaluations of self made by other persons. Journal of Personality, 19S7, 25, 393-
411. Heider, F. The psychology of interpersonal relations. New York: Wiley, 1958. Johnson, R., Conlee, M.,
& Tesser, A. Effects ol similarity of fate on bad news transmission: A reexamination. Journal of Personality
and Social Psychology, 1974, 29, 644-648. Langer, E. J., & Abelson, R. P. The semantics of asking a favor:
How to succeed in getting help without really dying. Journal of Personality and Social Psychology, 1972,
24, 26-32. Manls, M., Cornell, S. D., & Moore, J. C. Transmission of attitude-relevant information through
a communication chain. Journal of Personality and Social Psychology, 1974, 30, 81-94. Rosen, S., & Tesser,
A. On the reluctance to communicate undesirable information. The MUM effect. Sociometry, 1970, 33,
2S3-263. Schneider, D. J. Tactical self-presentation after success and failure. Journal of Personality end
Social Psychology, 1969, 13, 262-268. Siegel, S. A''onparametric statistics for the behavioral sciences. New
York- McGraw-Hill, 1956. Tesser, A., Rosen, S., & Batchelor, T. On the reluctance to communicate bad
news (the MUM effect): A role play extension. Journal of Personality, 1972, 40, 88-103. Tesser, A., Rosen,
S., & Tesser, M. On the reluctance to communicate undesirable messages (the MUM effect): A field study.
Psychological Reports, 1971, 29, 651-654. (Received October 14, 1975)